What happened
International Business Times reports that Broadcom is discussing a debt financing package of roughly $70 billion to $80 billion for chips and computing infrastructure tied to AI customers including Anthropic. The report says the proposed financing could include about $45 billion in senior debt and $35 billion in junior debt, while Bloomberg reported that the overall transaction could eventually reach $100 billion.
International Business Times reports that Broadcom is in talks to raise approximately $70 billion to $80 billion in debt to finance chips and computing infrastructure for artificial intelligence companies, including Anthropic. The report attributes the proposed structure to CNBC reporter David Faber and says the figures remain fluid while negotiations continue. Because the source describes discussions rather than a completed transaction, the financing should be treated as proposed, not committed or closed. That distinction remains important while the reported participants and terms are still being discussed.
According to IBTimes, the proposed package would be divided into multiple layers of debt. A senior tranche, which would have priority for repayment, is expected to total roughly $45 billion, while a junior tranche could reach about $35 billion. The article says Blackstone and Apollo Global Management are among the investment firms discussing participation. It also reports that Bloomberg, which first reported the talks, said the overall transaction could ultimately reach $100 billion. The different figures describe a package whose size and composition may evolve as negotiations proceed.
IBTimes places the discussions within Broadcom’s wider AI infrastructure strategy. The article says Broadcom announced a platform in June intended to enable 20 gigawatts of computing capacity for customers including Anthropic and OpenAI, and that Blackstone and Apollo led an initial $35 billion financing connected to that initiative. These details are reported by IBTimes; the source does not provide a new Broadcom filing or financing agreement independently confirming the latest proposed debt package. The earlier financing and the latest discussions should therefore be understood as related reported developments, not as one confirmed final transaction.
Read the primary source: ibtimes.com ↗
Why it matters
The reported financing would show how AI infrastructure is increasingly being funded through large, layered debt transactions rather than only through technology-company cash and equity. It also illustrates the growing role of chipmakers and infrastructure suppliers in arranging capital for the data centers and computing systems needed to train and run AI models.
If completed, a financing package of this scale would demonstrate how the AI buildout is becoming a major infrastructure-finance market. The money would support physical assets such as computing equipment, data centers, electricity capacity and networking systems. IBTimes says the sums involved are beginning to resemble those associated with major infrastructure projects rather than conventional technology investments. That comparison highlights the scale of the capital required to expand AI capacity and the financial structures being considered to provide it.
The report also describes a shift in the role of chip companies. According to IBTimes, suppliers are increasingly helping assemble the capital needed to buy processors and build the facilities where those processors will operate. Broadcom’s custom AI accelerators and networking products position it as an important supplier for hyperscalers and AI developers seeking alternatives or complements to Nvidia’s graphics processors, although the article does not provide independent performance or market-share data. The reported arrangement consequently links the sale of hardware with the financing of the infrastructure that uses it.
The financial exposure is substantial. IBTimes notes that data centers require very large upfront investments, while advances in semiconductor technology can reduce the useful life of expensive equipment. Investors therefore must weigh the durability of AI demand, the ability of customers to generate revenue from new systems, and the risk that infrastructure is built faster than utilization develops. The source does not establish that the proposed financing is economically sound or that the underlying AI demand will meet expectations. Those unresolved questions are central to how lenders and investors would assess the reported debt.
What to watch next
The discussions have not been finalized, and the reported amounts remain subject to change. Key developments include whether Broadcom completes the financing, how much debt is ultimately raised, which AI customers and projects support repayment, and whether investors continue accepting long-term exposure to rapidly changing AI hardware and uncertain AI revenue.
The immediate question is whether the talks become a completed transaction and whether the final amount resembles the reported $70 billion to $80 billion range. The senior and junior tranches could change in size, pricing, structure or participants before any agreement is announced. IBTimes explicitly says the figures remain fluid, so the report does not establish final borrowing terms. A public announcement or other formal documentation would be needed to determine whether the proposed structure has been completed.
Further documentation would be needed to clarify which Broadcom assets, facilities or customer commitments would support repayment. The source names Anthropic and OpenAI as customers connected to Broadcom’s broader platform, but it does not say that either company has guaranteed the proposed debt or accepted responsibility for it. It also does not identify the exact facilities, locations, schedules or contractual arrangements covered by the latest discussions. These details would help distinguish general customer relationships from obligations directly connected to the financing.
The broader market signal will be whether other chipmakers and infrastructure providers pursue similar financing models. IBTimes reports that Nvidia said it would provide up to $105 billion in financing for a new OpenAI data center in Ohio and had partnered with major asset managers on a separate $500 billion infrastructure initiative. Those figures are included as context from the source, not as independent confirmation here. Readers should watch for securities filings, lender disclosures, customer commitments and evidence of actual construction or deployment before treating the reported financing as settled. Such evidence would also show how the reported plans translate into operating infrastructure.
